Financial Datasets @findatasets

I’m rebuilding my segmented financials API. Hit a wall when researching stocks last night. Why: I need my AI agents to see exactly how a company makes money. So I’m adding: • revenue by geography • product line • business segment • and more Once that’s done, I’ll add…

Today we crossed 41,000 stars ✨ What started as a tiny repo to test some ideas is growing into a small movement. An open-source hedge fund stack that anyone can build on. Been excellent working with LangGraph and @LangChainAI - the foundations of this project.

I’m exploring next steps for our AI hedge fund. Goal: agents that find anomalies in noisy, divergent market data. Starting with: • news sentiment agent • earnings quality agent • bankruptcy risk agent They’ll mix rule-based logic with LLMs. An arsenal of agents that spot…

It’s finally live. SEC filings on the @findatasets MCP server. Features: • Pull text from filings • Query by section (Item) • Search across millions of filings Now you can mix qual + quant data for deeper research. No more parsing SEC filings yourself. The raw text is…

Introducing the Analyst Estimates API 🔮 I built this for 3 months and it’s finally live. It includes: • earnings per share • forecasts for 3 years • nightly updates Estimates are generated by internal models + data from @findatasets. Probabilistic, not guarantees. Avg…

I built an email notifier for SEC filings. The second earnings drop, I get email with: • financials • key metrics • earnings misses Next idea: run LLMs on filings to extract insights, flag new risks, and more. Do you want this? If enough interest, I’ll launch and make it…

Stock market data shouldn’t cost $20,000. I built @findatasets to fix this. For $20 you get: • 30 years of earnings • Every S&P 500 company • In under 30 secs No contracts. No restrictions. No limits.

I’m working on Background Agents today. Think of them as always-on helpers for stock research. Key Features: • Monitor markets → never miss news • Analyze in parallel → track more tickers • Send alerts → catch signals in real time With @LangChainAI, we can add…

I added benchmarks to our backtester. Now our AI agents compete with the market. Benchmark examples: • S&P 500 • Nasdaq • MSCI World No more guessing if our strategy is doing well. The comparison is right there. One more step toward a true open source hedge fund.

It’s easy to underestimate AI agents. You think: they’ll crunch numbers, do some some simple tasks, maybe speed up research. But then you see them rebalance portfolios, run market simulations, and nail calls humans miss. This isn’t hype. It’s real and happening now.

I just added more evals to our AI hedge fund. We use LLMs, so evaluation is everything. Now we test: • dollar cost averaging • portfolio rebalancing • multi-entry / exit scenarios Across long and short strategies. The foundation is set. Now the fun begins.

I’m finally mastering MCP + OAuth this week. OAuth has been a massive pain point for me. My goal: make it dead simple for anyone to add auth to their own MCP server. I’ll share learnings here on X.

I’m more convinced of this every day: The future of investing will be powered by AI. We won’t hand our savings to money managers charging fees. AI agents will invest for us, for free. And they’ll crush it.

I just upgraded our Charlie Munger agent. This one is my favorite. Combines 3 things: • determinism (moat, mgmt, valuation) • judgment (LLM signals) • mental models (invert, margin of safety) At its core, prioritizes wonderful businesses at fair prices.

Today we crossed 40,000 stars ✨ It all started as a simple idea, riding the NYC subway early in the morning, otw to work. The vision: an open-source, multi-agent hedge fund that could rival Wall Street. Still tons to do, but it’s been a joy building with @LangChainAI,…

I finally cracked the next stage of scale. The @findatasets Developer tier now handles 1000 requests/min (up from 200). Behind the scenes: • optimized DB indexes • smarter cache invalidation • finetuned connection pooling p95 latency still <200ms. Perfect for multi-agent…

The future of finance isn’t closed. It’s open source. We’re building an AI hedge fund in public. Transparent. Collective. Owned by us all.

I just upgraded our Warren Buffett agent. It’s now 89% faster at generating signals. LLM changes: • reduce prompt by 56% (more focus) • lower task complexity (less work) • max output at 100 tokens (less generation) The agent went from taking ~29 secs to ~3 secs. Simple…

I’m excited for this. Analyst estimates on @findatasets. Two things I’m working on: 1 • A hosted API 2 • An open-source ML model One of the most requested features from people building AI agents. My top priority this week.

Great to see the MCP ecosystem grow. Our @findatasets server usage is up 7x since early Aug. Top demand: real-time data • stock prices • crypto prices • earnings + financials Plenty of work still ahead, especially around MCP ↔ authentication. But I’m bullish: protocols…

OpenAI Developers @OpenAIDevs

3 weeks ago

Great to see the MCP ecosystem grow. Our @findatasets server usage is up 7x since early Aug. Top demand: real-time data • stock prices • crypto prices • earnings + financials Plenty of work still ahead, especially around MCP ↔ authentication. But I’m bullish: protocols…

269 733 5K 988K 2K

Download Video